This is the “UNFOLLOW” tweet. If you see this and you don’t remember why did you even follow me in the first place - press three dots then “Unfollow @ivanov_dev”.

My secret sauce when working with Claude Code - it’s called *thinking*

Is there any reason for this at all?. Feels like it's done out of spite.

What an irony - a required voluntary self-identification

I'm excited to host a YT stream this Wednesday!. We'll discuss the common issues in today's programming courses and then, I'll unveil a new teaching approach. Can't wait to share!.

Facebook ads manager, I'm not sure why they stopped at three scollbars

Lesson learned. NEVER post tiktoks from the browser, it's a sure way to get 0 views. Here, check this out. *the video with a few thousand views had paid promotion, ignore it

Ok, new challenge, I upload a funny video to TikTok every day until I don't.

I decided to post a TikTok about programming every day for 30 days to see what happens:. I'm webdevivanov there 👇

3/3 Grab the first chapter for free here:

2/3 The goal of this book is to make confusing concepts clear. Some examples:.- what is HEAD.- how to reset/revert/restore and what's the difference between those actions.- what are stashes and how to not get lost with them.

1/3 I wrote a book about using Git from command line. But there are tons of books about Git? How's yours different?. Mine focuses on ✨imagination ✨. No, seriously. You can save a ton of confusion if you imagine things a certain way.

Password managers should allow generating a random password in ANY input. I won’t be upset if i see an option to “Generate strong password” in a textarea . For some reason most password managers are very picky about the input fields.

RT @newlinedotco: 𝐐𝐮𝐢𝐜𝐤𝐥𝐲 𝐃𝐞𝐩𝐥𝐨𝐲 𝐍𝐞𝐱𝐭.𝐣𝐬 𝐓𝐨 𝐕𝐞𝐫𝐜𝐞𝐥. In this short tutorial, Maksim (@ivanov_dev) demonstrates how to deploy a #Next.js appl….

As an alternative to AI alignment we could develop AI immunity . To do this we should create an AI vaccine. Create malevolent agents while our AI technologies are still weak and let them loose. This will give us a chance to prepare for more dangerous forms of AI.

Are we certain that the models that we have now are not enough to cause damage?. What if we take an available model, DAN it feed in some interesting information, like some city electric infrastructure or water supply plans?. Would it be able to find some good attack vectors?.

What if Yudkovski is influenced by some AI that wants to win in a darwinian AI competition?.

Swipe left on iOS:.- opens a nav panel .- opens camera.- archives a message.- selects a message.- opens a context menu.- switches between tabs.- does nothing. What did I miss?.

Hallucinations are not unique to LLMs. Humans do this all the time. Good example: split brain experiment. Common example: all the self-help gurus out there.

iOS triple-tap screenshot feature works once for every 10 attempts. WAIDW?.

Stupid ducking iOS keyboard. Learn some new words, look at all those LLMs out there.